Identification Method Of Porosity In Weld Seam Image Based On Optimization Algorithm

With the continuous improvement of automation,scale and refinement in modern industrial production,automatic identification technology has gradually become an important development direction of intelligent manufacturing.NDT plays the core role in safety assessment of welding quality,but weld X-ray film the traditional manual distinguishing method is inefficient,subjective,has been difficult to meet the needs of machine production and quality information..Detection and analysis of online using X ray NDT computer aided evaluation system can effectively overcome the artificial assessment because of eye fatigue,experience caused by omissions and false positives,so as to effectively solve the efficiency and evaluation of artificial detection problem is not a.In recent years,the use of machine vision for radiographic nondestructive testing has become the focus of research at home and abroad,and the most critical technology is the automatic recognition of ray images based on machine vision.In this paper,the method and technology of automatic flaw identification based on X ray image of welding seam are studied.The industrial camera is used to shoot or scan the ray image,and the digitization of the weld line image is realized.An adaptive method for detecting the porosity of weld radiographic images is developed by using digital image processing technology.Compare the advantages and disadvantages of the commonly used image edge extraction algorithm in the field of weld defect detection,in-depth analysis of the manufacturing of pressure vessel welding porosity formation and X ray image features,the stomata identification method based on gray gradient,considering the different direction of the gray gradient algorithm,enhance the sensitivity of the hypotenuse and circular contour.The adhesion of pores and small pores of the false detections,the first weld location under complex background using the local dynamic threshold method,and then design a multi direction template for Sobel edge detection and region labeling,finally analysis of gray gradient in the marked area of the original image X and Y directions,effectively enhance the small pores and stomata adhesion recognition ability.The test results show that the algorithm has good accuracy,adaptability and rapidity in practical application,and lays a foundation for subsequent statistical analysis of holes and welding quality evaluation.Compared with the traditional manual evaluation method,the proposed method based on machine vision has advantages in four aspects,such as the discovery of hole defects,the counting of statistics,the measurement of dimensional accuracy and the accumulation of time.In comparison with other automatic identification systems,the method proposed in this paper has some improvements in three aspects,such as small hole recognition,adhesion,hole recognition,and pore size accuracy.